Estimation of treatment effects in matched-pair cluster randomized trials by calibrating covariate imbalance between clusters

Research output: Contribution to journalArticle

Abstract

Summary: We address estimation of intervention effects in experimental designs in which (a) interventions are assigned at the cluster level; (b) clusters are selected to form pairs, matched on observed characteristics; and (c) intervention is assigned to one cluster at random within each pair. One goal of policy interest is to estimate the average outcome if all clusters in all pairs are assigned control versus if all clusters in all pairs are assigned to intervention. In such designs, inference that ignores individual level covariates can be imprecise because cluster-level assignment can leave substantial imbalance in the covariate distribution between experimental arms within each pair. However, most existing methods that adjust for covariates have estimands that are not of policy interest. We propose a methodology that explicitly balances the observed covariates among clusters in a pair, and retains the original estimand of interest. We demonstrate our approach through the evaluation of the Guided Care program.

Original languageEnglish (US)
Pages (from-to)1014-1022
Number of pages9
JournalBiometrics
Volume70
Issue number4
DOIs
StatePublished - Dec 1 2014

Fingerprint

Matched pairs
Randomized Trial
Treatment Effects
Design of experiments
Covariates
Research Design
experimental design
methodology
Experimental design
Assignment
Methodology
Evaluation
Estimate
Demonstrate

Keywords

  • Bias correction
  • Causality
  • Covariate-calibrated estimation
  • Guided Care program
  • Meta-analysis
  • Paired cluster randomized design
  • Potential outcomes

ASJC Scopus subject areas

  • Applied Mathematics
  • Statistics and Probability
  • Agricultural and Biological Sciences(all)
  • Biochemistry, Genetics and Molecular Biology(all)
  • Immunology and Microbiology(all)
  • Medicine(all)

Cite this

@article{b657f2fe078346fcad93ef9988887387,
title = "Estimation of treatment effects in matched-pair cluster randomized trials by calibrating covariate imbalance between clusters",
abstract = "Summary: We address estimation of intervention effects in experimental designs in which (a) interventions are assigned at the cluster level; (b) clusters are selected to form pairs, matched on observed characteristics; and (c) intervention is assigned to one cluster at random within each pair. One goal of policy interest is to estimate the average outcome if all clusters in all pairs are assigned control versus if all clusters in all pairs are assigned to intervention. In such designs, inference that ignores individual level covariates can be imprecise because cluster-level assignment can leave substantial imbalance in the covariate distribution between experimental arms within each pair. However, most existing methods that adjust for covariates have estimands that are not of policy interest. We propose a methodology that explicitly balances the observed covariates among clusters in a pair, and retains the original estimand of interest. We demonstrate our approach through the evaluation of the Guided Care program.",
keywords = "Bias correction, Causality, Covariate-calibrated estimation, Guided Care program, Meta-analysis, Paired cluster randomized design, Potential outcomes",
author = "Zhenke Wu and Constantine Frangakis and Thomas Louis and Scharfstein, {Daniel O}",
year = "2014",
month = "12",
day = "1",
doi = "10.1111/biom.12214",
language = "English (US)",
volume = "70",
pages = "1014--1022",
journal = "Biometrics",
issn = "0006-341X",
publisher = "Wiley-Blackwell",
number = "4",

}

TY - JOUR

T1 - Estimation of treatment effects in matched-pair cluster randomized trials by calibrating covariate imbalance between clusters

AU - Wu, Zhenke

AU - Frangakis, Constantine

AU - Louis, Thomas

AU - Scharfstein, Daniel O

PY - 2014/12/1

Y1 - 2014/12/1

N2 - Summary: We address estimation of intervention effects in experimental designs in which (a) interventions are assigned at the cluster level; (b) clusters are selected to form pairs, matched on observed characteristics; and (c) intervention is assigned to one cluster at random within each pair. One goal of policy interest is to estimate the average outcome if all clusters in all pairs are assigned control versus if all clusters in all pairs are assigned to intervention. In such designs, inference that ignores individual level covariates can be imprecise because cluster-level assignment can leave substantial imbalance in the covariate distribution between experimental arms within each pair. However, most existing methods that adjust for covariates have estimands that are not of policy interest. We propose a methodology that explicitly balances the observed covariates among clusters in a pair, and retains the original estimand of interest. We demonstrate our approach through the evaluation of the Guided Care program.

AB - Summary: We address estimation of intervention effects in experimental designs in which (a) interventions are assigned at the cluster level; (b) clusters are selected to form pairs, matched on observed characteristics; and (c) intervention is assigned to one cluster at random within each pair. One goal of policy interest is to estimate the average outcome if all clusters in all pairs are assigned control versus if all clusters in all pairs are assigned to intervention. In such designs, inference that ignores individual level covariates can be imprecise because cluster-level assignment can leave substantial imbalance in the covariate distribution between experimental arms within each pair. However, most existing methods that adjust for covariates have estimands that are not of policy interest. We propose a methodology that explicitly balances the observed covariates among clusters in a pair, and retains the original estimand of interest. We demonstrate our approach through the evaluation of the Guided Care program.

KW - Bias correction

KW - Causality

KW - Covariate-calibrated estimation

KW - Guided Care program

KW - Meta-analysis

KW - Paired cluster randomized design

KW - Potential outcomes

UR - http://www.scopus.com/inward/record.url?scp=84919674731&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84919674731&partnerID=8YFLogxK

U2 - 10.1111/biom.12214

DO - 10.1111/biom.12214

M3 - Article

C2 - 25163648

AN - SCOPUS:84919674731

VL - 70

SP - 1014

EP - 1022

JO - Biometrics

JF - Biometrics

SN - 0006-341X

IS - 4

ER -